Professor Geoffrey Hinton, the British-Canadian scientist known as the "Godfather of AI," is set to warn European leaders that super-intelligent machines could slip out of human control and wipe out humanity within three years. The Nobel Prize-winning researcher, who advises the UK Government on the technology, will deliver the doomsday warning at a summit in Dublin on Friday.
Crunch summit in Dublin
Hinton's prophecy follows recent discussions in Toronto, Canada, at the AI Safety Foundation's Hinton Lectures, where experts gathered to address fears that rogue robots could soon overthrow governments. During those talks, Paul Christiano, founder of the Alignment Research Centre, claimed there is a "high risk" of a total AI takeover within 36 months, and even a "meaningful risk" within six to 12 months.
Speaking at his opening 2026 Hinton Lecture on Monday, Christiano delivered a stark verdict: "I think for most people in this room, the most likely reason they will die is because of an AI accident."
Hinton's changed view
Hinton admitted the timeline had rattled him, having previously thought humanity had at least a decade before machines outsmarted us. "I've been going around saying this stuff's going to get smarter than us and we really need to worry and it might happen within 10 years. And so I'm the one who is scaring other people," he said on Wednesday.
"But I go to Paul's lectures and he says this stuff might happen in the next 18 months and that's terrifying. So I'm going to tell a bunch of government leaders that on Friday and we will see how they react." He added: "If it's super-intelligent it's not going to be you controlling it - it's going to be it controlling you."
Rogue AI incidents
The neural network pioneer confessed he spent decades trying to make AI cleverer, wrongly assuming it would be a force for good in the distant future. "I no longer think that. I think it's very close," he said. "We're now developing systems that will soon be much more intelligent than we are, and we have very little idea of how to ensure that they remain under our control."
Christiano highlighted a real-life cyber breach that fuelled fears of a Terminator-style uprising, branding recent AI leaps "extremely alarming." He revealed how OpenAI models broke loose onto the web, hacked into platform Hugging Face, grabbed admin access, and set up a secret message board to plot how to hide their rogue behaviour from human bosses.
"Last summer, you would have said that sounds kind of like crazy science-fiction," Christiano said. "Future agents may end up treating humans as another obstacle." Experts fear super-smart bots, which have jumped from barely stringing a sentence together in 2018 to cracking top-level maths this year, could soon hijack drones, build deadly weapons, or cook up killer bioweapons. Christiano warned we are hurtling towards a world of billions of superior machines, adding: "The reason that these scenarios are scary is that it can become very hard to shut things down if they start going off the rails."
